IP connectivity delivers flexibility and scalability but making the theory work often requires integrated solutions that are adaptable, open, and promote interconnectivity.
These challenges are further compounded when we introduce the concepts of managed and unmanaged IP networks, especially as the public internet is becoming increasingly utilized.
Traditional broadcast workflows focused on contribution and distribution often had the advantage of point-to-point connectivity that guaranteed bandwidth, latency, and redundancy.
As we progress to IP, and with it the flexibility and scalability that is provided, the attributes of the static SDI/AES and analog circuits can no longer be taken for granted. Instead, we must look at IP more from the IT perspective to deliver the promised advantages.
IP connectivity is available as managed or unmanaged services. That is, bandwidth, latency, and security are guaranteed to varying levels depending on the service level agreement with the supplier contract. This requires broadcasters to think laterally about how the various attributes and parameters of how the contribution circuits are provisioned.
And the main requirements of many broadcasters are flexibility, low latency, and reliability. Furthermore, security is playing an increasingly important role for unmanaged services.
The high barrier to entry for traditional broadcast systems often resulted in high levels of security being implemented by default. For example, the cost of a VT machines was a barrier for most casual criminals as playing a Digibeta tape often relied on the procurement of a $50K machine and the expertise to go with it. However, in the IP world, the cost of entry for cybercriminals is much lower. Consequently, we need to protect high value media even more.
The media flow as a whole can be encrypted using systems such as AES (Advanced Encryption Standard) and BISS (Basic Interoperable Scrambling System). Both AES and BISS encrypt the video and audio flows directly. This has the advantage of reducing the risk of anybody sniffing and accessing the content but does mean users needing access to the content must be in possession of the relevant keys.
AES is a generic encryption system and uses a symmetric key encryption meaning that both the encoder and decoder use the same key. This has the advantage of being faster and more resource efficient than asymmetric key encryption, but one of the drawbacks is that secure methods of key management must be adopted. Furthermore, new keys must be regularly generated in case one of the key users inadvertently misplaces the key.
BISS2 was developed by the EBU specifically for broadcasters and four modes are specified: Mode 0, Mode 1, Mode E and Mode CA. They vary in their complexity and security depending on their application. For example, Mode 1 was designed specifically for DSNG, fly-away, and emergency type applications and is the fallback mode for BISS2 compliant media exchange.
A 32-character number is shared by the sender and receiver, known as the Session Word and is manually entered into the encoder and decoder, allowing the media flow to be encrypted. Although this provides a good level of encryption for the media flow, Mode-CA takes this a stage further by encrypting the keys. It uses both symmetrical and asymmetrical keys to combine a complex key encryption and SW exchange system to improve secure exchange of keys and allow an encoder to target specific decoders. Thus, providing a high level of media protection.
One of the fundamental strengths of IP is that the data packets are hardware agnostic. That is, they have no “knowledge” of the type of hardware infrastructure they are being transferred over. This could be ethernet or fiber, or RF and WiFi. The ease with which IP packets can be routed between different infrastructures further adds to its strengths.
There are flags within the IP header that indicate the higher level protocols the IP packet is aligned to, such as UDP or TCP, but in the most part, the actual data being caried is independent of the actual IP data section. Again, this further adds to the flexibility that IP has to offer as we can transfer any type of data we like, whether its control, data files, or streamed media.
The power of IP has laid open many opportunities for broadcasters, including the provision of IP services for contribution services used in OBs. Traditionally, OBs have used dedicated SDI and AES circuits provided by Telcos, satellite and RF links. Each of these has their own challenges with lack of flexibility and high costs being at the top of the list. IP services, often provided by Telcos are more flexible and lower in cost than the traditional methods of delivery.
IP And QoS
One of the challenges of the provision of IP services is that they can be either managed or unmanaged. A managed service provides compliance with a much more tightly specified QoS (Quality of Service). This includes packet loss, bit rate, data throughput, latency, and jitter. However, managed services are not always available so a broadcaster may have to work with the unmanaged service instead.
The QoS metrics are important as attributes such as packet loss, latency, and jitter can have a massive effect on the QoE (Quality of Experience) for the viewer. Picture freezing, video dropout, and audio distortion can all have a significant impact on the QoE leading to viewer complaints, resulting in them switching over to another service.
Figure 1 – QoS metrics for IT networks covers many parameters that broadcasters have traditionally taken for granted when working in SDI and AES networks.
Packet loss not only leads to potential picture break up for video and audio distortion but can have a disproportionate effect on control and monitoring. Although video and audio tend to transfer over contribution networks using UDP/IP, that is fire-and-forget, control and monitoring will use TCP/IP connectivity to guarantee delivery of monitoring and control data. Dropped or delayed packets can initiate the resend-timeout feature leading to large latency occurring between the control and monitoring devices.
One of the unintended consequences of resend-timeout in TCP/IP is that the data rate can appear to be high, as the lost packets are being resent, but the overall data throughput is very low. For any broadcast engineer to effectively utilize IP contribution services they must understand the difference between data rate of packets on the wire, and data throughput provided by protocols such as TCP/IP. Often, they are quite different, which again is a massive difference from how we worked with SDI and AES services.
TCP/IP is an adaption of the ARQ (Automatic Repeat Query) strategy that provides error control in lossy networks such as the internet. ARQ uses UDP/IP packets to exchange data and resend any packets that are lost. Although TCP adds congestion control in addition to ARQs error correction, similar latencies are apparent in ARQ. However, if ARQ is used as part of a custom or proprietary solution, the vendors can tune the ARQ parts of the algorithm to specific applications. This allows them to better stream media and potentially improve on latency.
As a rule of thumb, the tighter the constraints within the QoE metrics, the more expensive the service will be. Although unmanaged services will be less expensive than managed, there is a potential cost associated with this in terms of packet dropout, latency, and jitter, as well as overall reliability. In the whole, it is possible to work with either and possibly both at the same time, but the benefits QoE metrics bring to the contribution network required for an OB must be well understood.
To overcome the QoE limitations of unmanaged services, some form of monitoring is required. This can either be provided manually using network analysis tools, or more productively using automated detection and change over codecs. For example, an OB may be using a managed service for the main contribution feed but an unmanaged service for its back up. An automated system will be able to constantly monitor the networks and switch over appropriately to achieve switchover should one of the services fail without manual intervention.
Unmanaged circuits certainly have their place, it’s just that we must be aware of some of the challenges we face when using them. Low level Packet loss and jitter can be overcome using ARQ, FEC (forward error correction) and buffers. However, if a sequence of video is delivered corrupted then the receiver either has to try and fix it, request a re-send, or just flag it as an error. This results in either increased latency or a loss of video quality. Packet jitter has similar challenges, but again this can be fixed with buffers. However, buffers introduce variable latency.
It’s important to note that packet dropout, latency, and jitter are a fact of life when working with IP services, even with managed circuits. However, what is important is the predictability of such systems. This is possible with managed networks but less so with unmanaged. It’s much easier to work with known and specified latencies and packet jitter within a system.
Although latency may be perceived as the enemy of broadcasters, what is more important is determining predictable latency. We can work with 100ms or 200ms of latency, within a few milliseconds of tolerance, what is very difficult to work with is a latency that violently swings between 100ms and 200ms.
It’s also worth remembering that we have suffered from dropout, latency, and jitter in television since we broadcast the first transmissions in the 1930s. It’s just that the tight timing constraints we’ve always worked with help keep these metrics so low we barely noticed them. Fast forwarding to the 21st century, there is an argument to suggest that the nanosecond timing developed for SDI is no longer really needed: we no longer use cathode ray tube cameras and televisions, so we don’t need to worry about frame accurate timing to within a few microseconds. Modern flat panel televisions and CCD/CMOS cameras are much better at dealing with timing and don’t need such tight tolerances.
When providing contribution from OBs we must consider the return path. Video streaming over UDP/IP can theoretically work without a return path, as the data just travels in one direction, the reality is that other applications within the network will be using the return path for ARQ and TCP/IP as well as reverse vision, sound, and IFBs. Again, keeping the latency within tight tolerances helps enormously.
You might also like...
Compression is almost taken for granted despite its incredible complexity. But it’s worth remembering how compression has developed so we can progress further.
John Watkinson moves on to discussion of the effects of the medium waves are travelling in and explains why loudspeaker enclosures contain foam.
Compression is the ultimate enabling technology behind broadcasting. Without it, life would be very difficult indeed. In this new series, the whole topic will be explored at some depth.
Broadcasting is totally dependent on waves which crop up in a surprising number of places. Sound waves and light waves form the message, which is delivered by further types of wave.
Delivering determinant latency is more important than fighting variable latency, even if it is small. In this article, we look at how codec design and JPEG-XS can scale to make the best use of network bandwidth while keeping latency predictable.